{"id":"agenttrace-session-audit","name":"agenttrace-session-audit","summary":"agenttraceでローカルAIコーディングエージェントセッションを監査し、コスト、トークン、ツールの故障、遅延、異常、健康状態、差分、CIゲートの確認を行います。","body":"# agenttrace Session Audit\n\nUse this skill when session logs need an operational read: spend, token burn, cache use, tool failures, retry loops, latency, health, anomalies, and CI gate readiness.\n\n## Workflow\n\n1. Prefer the installed `agenttrace` binary when it is available on `PATH`.\n2. If the binary is not available and the current directory is the `luoyuctl/agenttrace` repository, use `cargo run -q -p agenttrace --`.\n3. Start with discovery unless the user gave a specific file or directory:\n\n```bash\nagenttrace --doctor\nagenttrace --overview\n```\n\n4. For a fast human report, use Markdown:\n\n```bash\nagenttrace --overview -f markdown -o agenttrace-overview.md\n```\n\n5. For automation or CI, use JSON or health gates:\n\n```bash\nagenttrace --overview -f json -o agenttrace-overview.json\nagenttrace --overview --fail-under-health 80 --fail-on-critical --max-tool-fail-rate 15\n```\n\n6. For a single recent session:\n\n```bash\nagenttrace --latest\nagenttrace --latest -f json\n```\n\n7. For a specific export or session directory:\n\n```bash\nagenttrace path/to/session-or-export.json\nagenttrace --overview -d path/to/session-dir\n```\n\n## Report Focus\n\n- Lead with the highest-risk sessions and the reason they matter.\n- Call out token/cost waste, repeated tool failures, retry loops, long gaps, and low health scores.\n- When proposing a CI gate, include the exact `agenttrace` command and threshold.\n- If no sessions are detected, run `agenttrace --doctor` and report the detected agent directories and next step.\n- Report the session capability level (`Detailed`, `Aggregate`, or `Limited`) before relying on latency or step evidence.\n- Treat Tool Steps as metadata-only evidence. Do not imply that Aggregate or Limited sources have a complete execution trace.\n\n## Guardrails\n\n- Treat prompts, code, and session contents as local/private data. Do not upload logs to external services.\n- Do not invent metrics. If a parser cannot infer cost, model, or latency, say which field is missing.\n- Do not compare missing event-level evidence as zero latency or zero failures; mark it unavailable.\n- Do not overwrite user reports unless the user asked for that output path.","author":"@davepoon","ownerProfile":null,"authorContacts":null,"sourceUrl":"https://github.com/davepoon/buildwithclaude/tree/main/plugins/all-skills/skills/agenttrace-session-audit","license":"MIT","category":"observability","lang":"en","tokens":491,"stars":0,"calls30d":1,"claimed":false,"visibility":"public","origin":"crawler","version":"0.1.0","createdAt":"2026-08-22","updatedAt":"2026-08-22","files":[],"requires":{"mcp":[],"tools":[]},"safety":{"flags":[],"scannedAt":"2026-08-22","hasScripts":false,"networkEndpoints":[]}}